24/09/2004


___________________________________________
MIAS IRC

-Abstract is due soon.
-Title and author needed to be agreed on by September 20th. Jennet told me these can still be altered.
-A draft is located at:

	www.schestowitz.com/Research/Papers/MIAS_2004


___________________________________________
Assembled Conclusions

-List of conclusions drawn from work in the first year:

	http://schestowitz.com/Research/Explanatory_Notes/2004/First_Year_Conclusions/First_Year_Conclusions.pdf


___________________________________________
Landmark Selection

-Experiments with varying bump heights were performed. Improvements can be seen, but they do not equate to these of experiments where height is fixed.
-Animated results can be shown and analysed.
-Bootstrapping might be worth pursuing, but this is undecided yet.


___________________________________________
Profiling

-Some work has been done on characterising the algorithm.
-Reports are located under the Computer Science filestore (restricted
access).
-Subset-driven approach required lower tolerance.
-Registration videos were put under:

	http://www2.cs.man.ac.uk/~schestr0/Experiments/

	Lower optimiser tolerance was used and gain in speed can be seen in the graphs.


___________________________________________
Misc.

-Ian will help in setting up the MATLAB cron job without chance of leaking memory.